def update_watchlist(self, user, since_tweet_id, page_not_found):
pre_watchlist_column = self.motor_column.pre_watchlist
watchlist_column = self.motor_column.watchlist
self.logger.debug(user)
doc = yield pre_watchlist_column.find_one({'user.id_str': user['id_str'], 'state': 1})
if not doc:
doc = yield pre_watchlist_column.find_one({'user.screen_name': user['screen_name'], 'state': 1})
if doc:
yield pre_watchlist_column.remove({'_id': doc['_id']})
del doc['_id']
else:
doc = yield watchlist_column.find_one({'user.id_str': user['id_str'], 'state': 1})
if not doc:
return
if 'campaign_ids' in doc['user']:
user.update({'campaign_ids': doc['user']['campaign_ids']})
doc['user'] = user
doc['since_tweet_id'] = since_tweet_id
doc['page_not_found'] = page_not_found
doc['updated_at'] = now_in_drnj_time()
doc['state'] = 0
yield watchlist_column.update({'user.id_str': doc['user']['id_str']}, doc, upsert=True)
def create_campaign(self, params):
campaigns_column = self.motor_column.campaigns
params.update({'created_at': now_in_drnj_time()})
user_id_strs_to_follow = str(params["user_id_strs_to_follow"])
user_screen_names_to_follow = str(params["user_screen_names_to_follow"])
# removing it to be used elsewhere
params.pop("user_id_strs_to_follow", None)
params.pop("user_screen_names_to_follow", None)
add_to_watchlist(params['campaign_id'], user_id_strs_to_follow, user_screen_names_to_follow)
yield campaigns_column.insert(params)
def create_campaign(self, params):
campaigns_column = self.motor_column.campaigns
params.update({'created_at': now_in_drnj_time()})
user_id_strs_to_follow = str(params["user_id_strs_to_follow"])
user_screen_names_to_follow = str(
params["user_screen_names_to_follow"])
# removing it to be used elsewhere
params.pop("user_id_strs_to_follow", None)
params.pop("user_screen_names_to_follow", None)
add_to_watchlist(params['campaign_id'], user_id_strs_to_follow,
user_screen_names_to_follow)
yield campaigns_column.insert(params)
def update_watchlist(self, user, since_tweet_id, page_not_found):
pre_watchlist_column = self.motor_column.pre_watchlist
watchlist_column = self.motor_column.watchlist
self.logger.debug(user)
doc = yield pre_watchlist_column.find_one({
'user.id_str': user['id_str'],
'state': 1
})
if not doc:
doc = yield pre_watchlist_column.find_one({
'user.screen_name':
user['screen_name'],
'state':
1
})
if doc:
yield pre_watchlist_column.remove({'_id': doc['_id']})
del doc['_id']
else:
doc = yield watchlist_column.find_one({
'user.id_str': user['id_str'],
'state': 1
})
if not doc:
return
if 'campaign_ids' in doc['user']:
user.update({'campaign_ids': doc['user']['campaign_ids']})
doc['user'] = user
doc['since_tweet_id'] = since_tweet_id
doc['page_not_found'] = page_not_found
doc['updated_at'] = now_in_drnj_time()
doc['state'] = 0
yield watchlist_column.update({'user.id_str': doc['user']['id_str']},
doc,
upsert=True)
def prepare_hist_and_plot(self, n_tweets, users, n_bins, campaign_id):
import numpy
import matplotlib.pyplot as plot
plot_graphs = False
hist = {
'user_creation': {
'data': None,
'bins': None,
},
'user_n_tweets': {
'data': None,
'bins': None,
},
'user_n_tweets_overall': {
'data': None,
'bins': None,
},
'n_tweets': None,
'n_unique_users': None,
'n_default_profile_image': None,
'n_lower_than_threshold': None,
}
self.logger.debug("How many tweets? %d" % n_tweets)
hist['n_tweets'] = n_tweets
# TODO: abort if there are more than 200000 tweets.
if n_tweets > 200000:
return
#
# How many unique users?
#
n_unique_users = len(users)
self.logger.debug("How many unique users? %d" % n_unique_users)
hist['n_unique_users'] = n_unique_users
######
sec_title = "Histogram of user creation dates?"
#
tmp_dates = []
for x in users:
tmp_date = x['user']['created_at']
if type(tmp_date) != float:
tmp_date = py_utc_time2drnj_time(tmp_date)
tmp_dates.append(tmp_date)
# tmp_dates = [py_utc_time2drnj_time(x['user']['created_at']) for x in users]
(hist['user_creation']['data'], hist['user_creation']['bins']) = numpy.histogram(tmp_dates, bins=n_bins)
if plot_graphs:
bins = hist['user_creation']['bins'][:-1]
width = (hist['user_creation']['bins'][1] - hist['user_creation']['bins'][0])/2
plot.bar(bins, hist['user_creation']['data'], width=width, align='center')
xticklabels = [time.strftime('%d %b %Y', time.gmtime(drnj_time2py_time(x))) for x in bins]
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('1.pdf', dpi=600)
#####
sec_title = "Histogram of number of tweets of each user in this campaign"
tmp_counts = [int(x['n_user_tweets']) for x in users]
#
(hist['user_n_tweets']['data'], hist['user_n_tweets']['bins']) = numpy.histogram(tmp_counts, bins=n_bins)
if plot_graphs:
bins = hist['user_n_tweets']['bins'][:-1]
data = hist['user_n_tweets']['data']
width = (hist['user_n_tweets']['bins'][1] - hist['user_n_tweets']['bins'][0])/2
plot.bar(bins, data, width=width, align='center')
xticklabels = bins
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('2.pdf', dpi=600)
#####
sec_title = "What percentage of them used the default profile image?"
#
n_default_profile_image = 0
for u in users:
if u['user']['default_profile_image']:
n_default_profile_image += 1
hist['n_default_profile_image'] = n_default_profile_image
self.logger.debug("%s: %0.2f%%" % (sec_title, 100*(float(n_default_profile_image)/n_unique_users)))
#####
sec_title = "Histogram of tweet counts of unique users"
tmp_counts = [int(x['user']['statuses_count']) for x in users]
(hist['user_n_tweets_overall']['data'],
hist['user_n_tweets_overall']['bins']) = numpy.histogram(tmp_counts, bins=n_bins)
if plot_graphs:
bins = hist['user_n_tweets_overall']['bins'][:-1]
data = hist['user_n_tweets_overall']['data']
width = (hist['user_n_tweets_overall']['bins'][1] - hist['user_n_tweets_overall']['bins'][0])/2
plot.bar(bins, data, width=width, align='center')
xticklabels = bins
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('3.pdf', dpi=600)
#
sec_title = "What percentage of them have lower than 5 tweets?"
n_lower_than_threshold = 0
for u in users:
if u['user']['statuses_count'] < 5:
n_lower_than_threshold += 1
hist['n_lower_than_threshold'] = n_lower_than_threshold
self.logger.debug("%s: %0.2f%%" % (sec_title, 100*(float(n_lower_than_threshold)/n_unique_users)))
self.logger.debug(hist)
# converting numpy.array's to normal python lists.
for k in hist.keys():
if type(hist[k]) == dict:
for k2 in hist[k].keys():
if type(hist[k][k2]) == type(numpy.array([])):
hist[k][k2] = list(hist[k][k2])
hist = {'campaign_id': campaign_id,
'histogram': hist,
'created_at': now_in_drnj_time()}
return hist
def store_friends_or_followers(self, user_id, IDS, drnjID, fof):
"""Stores/updates list of drenaj user data using raw twitter data
IDS -- list of user ids client crawler reports
"""
# db = mongo_client[DRENAJ_DB[DRENAJ_APP_ENVIRONMENT]]
queue_collection = self.application.db.motor_column.queue
graph_collection = self.application.db.motor_column.graph
num_new_discovered_users = 0
num_edges_inserted = 0
dt = now_in_drnj_time()
queue_query = {"id": user_id}
print queue_query
# ATC: This mechanism requires finding the id twice
# With indexing, this may not be a big problem
# Alternative is trying to update and catching the pymongo.errors.OperationFailure exception
n_ids = yield queue_collection.find(queue_query).count()
id_exists = n_ids > 0
if id_exists:
print 'ID EXISTS'
queue_document = {
"$set": {
fof + "_retrieved_at": dt,
"retrieved_by": drnjID
}
}
# creates entry if query does not exist
# queue_collection.update(queue_query, queue_document, upsert=True)
yield queue_collection.update(queue_query, queue_document)
else:
print 'ID NOT EXISTS'
if fof == 'friends':
queue_document = validate_document(
new_queue_document(), {
"id": user_id,
"id_str": str(user_id),
"profile_retrieved_at": 0,
"friends_retrieved_at": dt,
"followers_retrieved_at": 0,
"retrieved_by": drnjID
})
elif fof == 'followers':
queue_document = validate_document(
new_queue_document(), {
"id": user_id,
"id_str": str(user_id),
"profile_retrieved_at": 0,
"friends_retrieved_at": 0,
"followers_retrieved_at": dt,
"retrieved_by": drnjID
})
print "QUEUE DOCUMENT: " + queue_document
yield queue_collection.insert(queue_document)
num_new_discovered_users += 1
# process each user id in IDS
for id in reversed(IDS):
# Insert the newly discovered id into the queue
# insert will be rejected if _id exists
queue_document = validate_document(
new_queue_document(), {
"id": id,
"id_str": str(id),
"profile_retrieved_at": 0,
"friends_retrieved_at": 0,
"followers_retrieved_at": 0,
"retrieved_by": drnjID
})
yield queue_collection.insert(queue_document)
num_new_discovered_users += 1
dt = now_in_drnj_time()
if fof == 'friends':
source = user_id
sink = id
elif fof == 'followers':
source = id
sink = user_id
else:
return
edge = yield graph_collection.find_one({
"id": source,
"friend_id": sink
})
if edge == None:
doc = validate_document(
new_graph_document(), {
'id': source,
'friend_id': sink,
'id_str': str(source),
'friend_id_str': str(sink),
'record_retrieved_at': dt,
"retrieved_by": drnjID
})
yield graph_collection.insert(doc)
num_edges_inserted += 1
# TODO: Handle unfollows: Find edges that no longer exist and move old record to graph_history and add unfollow record
raise Return({
'num_new_users': num_new_discovered_users,
'num_new_edges': num_edges_inserted
})
def store_multiple_profiles(self, ids, S, drnjID, campaign_id):
"""
"""
# print "Received recent profile of ", v['name'], ' a.k.a. ', v['screen_name']
db = self.application.db
queue_coll = db.motor_column.queue
print S
for i in range(len(S)):
status = None
if 'status' in S[i]:
status = S[i]['status']
del S[i]['status']
else:
status = {}
status['text'] = None
status['user'] = S[i]
status['user']['history'] = False
DB_TEST_VERSION = 0.2
tweet_dat = validate_document(new_tweet_template(), {
"tweet": status,
# TODO: Replace this DB_TEST_VERSION with source code
# version later
"drenaj_service_version": DB_TEST_VERSION,
"campaign_id": campaign_id,
"record_retrieved_at": drnj_time.now_in_drnj_time(),
"retrieved_by": drnjID,
}, fail=False)
print tweet_dat
user_id = S[i]['id_str']
# print profile_dat
# Check Queue
now = drnj_time.now_in_drnj_time()
queue_query = {"id": user_id}
queue_document = validate_document(new_queue_document(), {
"id": int(user_id),
"id_str": user_id,
"profile_retrieved_at": now,
"$setOnInsert": {
"friends_retrieved_at": 0,
"followers_retrieved_at": 0,
},
"retrieved_by": drnjID
})
# creates entry if query does not exist
yield queue_coll.update(queue_query, queue_document, upsert=True)
# Insert to profiles
## profiles_query = {"profile.id": user_id}
## prof = profiles_collection.find_and_modify(profiles_query, remove=True)
## if prof is not None:
## profiles_history_collection.insert(prof)
##
## profiles_collection.insert(profile_dat)
# this call marks the current entries as history
# maybe we won't need this for certain queries
db.move_to_history(user_id)
db.insert_tweet(tweet_dat)
# tweets_collection.insert(tweet_dat)
ids.remove(int(user_id))
raise Return(ids)
def store_multiple_profiles(self, ids, S, drnjID, campaign_id):
"""
"""
# print "Received recent profile of ", v['name'], ' a.k.a. ', v['screen_name']
db = self.application.db
queue_coll = db.motor_column.queue
print S
for i in range(len(S)):
status = None
if 'status' in S[i]:
status = S[i]['status']
del S[i]['status']
else:
status = {}
status['text'] = None
status['user'] = S[i]
status['user']['history'] = False
DB_TEST_VERSION = 0.2
tweet_dat = validate_document(
new_tweet_template(),
{
"tweet": status,
# TODO: Replace this DB_TEST_VERSION with source code
# version later
"drenaj_service_version": DB_TEST_VERSION,
"campaign_id": campaign_id,
"record_retrieved_at": drnj_time.now_in_drnj_time(),
"retrieved_by": drnjID,
},
fail=False)
print tweet_dat
user_id = S[i]['id_str']
# print profile_dat
# Check Queue
now = drnj_time.now_in_drnj_time()
queue_query = {"id": user_id}
queue_document = validate_document(
new_queue_document(), {
"id": int(user_id),
"id_str": user_id,
"profile_retrieved_at": now,
"$setOnInsert": {
"friends_retrieved_at": 0,
"followers_retrieved_at": 0,
},
"retrieved_by": drnjID
})
# creates entry if query does not exist
yield queue_coll.update(queue_query, queue_document, upsert=True)
# Insert to profiles
## profiles_query = {"profile.id": user_id}
## prof = profiles_collection.find_and_modify(profiles_query, remove=True)
## if prof is not None:
## profiles_history_collection.insert(prof)
##
## profiles_collection.insert(profile_dat)
# this call marks the current entries as history
# maybe we won't need this for certain queries
db.move_to_history(user_id)
db.insert_tweet(tweet_dat)
# tweets_collection.insert(tweet_dat)
ids.remove(int(user_id))
raise Return(ids)
def store_friends_or_followers(self, user_id, IDS, drnjID, fof):
"""Stores/updates list of drenaj user data using raw twitter data
IDS -- list of user ids client crawler reports
"""
# db = mongo_client[DRENAJ_DB[DRENAJ_APP_ENVIRONMENT]]
queue_collection = self.application.db.motor_column.queue
graph_collection = self.application.db.motor_column.graph
num_new_discovered_users = 0
num_edges_inserted = 0
dt = now_in_drnj_time()
queue_query = {"id": user_id}
print queue_query
# ATC: This mechanism requires finding the id twice
# With indexing, this may not be a big problem
# Alternative is trying to update and catching the pymongo.errors.OperationFailure exception
n_ids = yield queue_collection.find(queue_query).count()
id_exists = n_ids > 0
if id_exists:
print 'ID EXISTS'
queue_document = {"$set":
{
fof + "_retrieved_at": dt,
"retrieved_by": drnjID}
}
# creates entry if query does not exist
# queue_collection.update(queue_query, queue_document, upsert=True)
yield queue_collection.update(queue_query, queue_document)
else:
print 'ID NOT EXISTS'
if fof == 'friends':
queue_document = validate_document(new_queue_document(), {
"id": user_id,
"id_str": str(user_id),
"profile_retrieved_at": 0,
"friends_retrieved_at": dt,
"followers_retrieved_at": 0,
"retrieved_by": drnjID
})
elif fof == 'followers':
queue_document = validate_document(new_queue_document(), {
"id": user_id,
"id_str": str(user_id),
"profile_retrieved_at": 0,
"friends_retrieved_at": 0,
"followers_retrieved_at": dt,
"retrieved_by": drnjID
})
print "QUEUE DOCUMENT: " + queue_document
yield queue_collection.insert(queue_document)
num_new_discovered_users += 1
# process each user id in IDS
for id in reversed(IDS):
# Insert the newly discovered id into the queue
# insert will be rejected if _id exists
queue_document = validate_document(new_queue_document(), {
"id": id,
"id_str": str(id),
"profile_retrieved_at": 0,
"friends_retrieved_at": 0,
"followers_retrieved_at": 0,
"retrieved_by": drnjID})
yield queue_collection.insert(queue_document)
num_new_discovered_users += 1
dt = now_in_drnj_time()
if fof == 'friends':
source = user_id
sink = id
elif fof == 'followers':
source = id
sink = user_id
else:
return
edge = yield graph_collection.find_one({"id": source, "friend_id": sink})
if edge == None:
doc = validate_document(new_graph_document(), {
'id': source,
'friend_id': sink,
'id_str': str(source),
'friend_id_str': str(sink),
'record_retrieved_at': dt,
"retrieved_by": drnjID
})
yield graph_collection.insert(doc)
num_edges_inserted += 1
# TODO: Handle unfollows: Find edges that no longer exist and move old record to graph_history and add unfollow record
raise Return(
{'num_new_users': num_new_discovered_users, 'num_new_edges': num_edges_inserted})
def post(self, *args, **keywords):
"""
`view`
- params:
- user_id: user_id of the twitter user that we want to
retrieve all tweets.
`store`
- params:
- tweet_data: array of tweets in json format.
`retweets`
- params:
- tweet_id: `id` of the tweet that we want to
retrieve all retweets.
`filter`
- not implemented yet.
"""
if len(args) > 1:
(action) = args
else:
action = args[0]
print action
verbose_response = self.get_argument('verbose', '')
if (action == 'view'):
try:
# TODO: we need to get here again.
user_id = self.get_argument('user_id')
# if no user_id is supplied.
if user_id == '':
tmp = []
else:
tweets_coll = self.application.db.motor_column.tweets
# running the query
cursor = tweets_coll.find({
'tweet.user.id_str': str(user_id),
})
tmp = [x for x in (yield cursor.to_list(length=100))]
self.write(bson.json_util.dumps({'results': tmp}))
self.add_header('Content-Type', 'application/json')
except MissingArgumentError as e:
# TODO: implement logging.
raise HTTPError(500, 'You didn''t supply %s as an argument' % e.arg_name)
elif (action == 'store'):
try:
tweet_data = self.get_argument('tweet_data')
campaign_id = self.get_argument('campaign_id', 'default')
watchlist_related = self.get_argument('watchlist_related', '')
if tweet_data:
#tweet_array = bson.json_util.loads(tweet_data, object_hook=self.datetime_hook)
tweet_array = bson.json_util.loads(tweet_data)
tmp_tweets = []
# TODO: Sanity check the data!
# For example, treat 'entities', 'user' specially.
DB_TEST_VERSION = 0.2
for tweet_obj in tweet_array:
tweet_obj['user']['history'] = False
tmp_tweets.append(validate_document(new_tweet_template(), {
"tweet": tweet_obj,
# TODO: Replace this DB_TEST_VERSION with source code
# version later
"drenaj_service_version": DB_TEST_VERSION,
# TODO: "retrieved_by": keywords['drnjID'],
"retrieved_by": "drenaj",
"campaign_id": campaign_id,
"record_retrieved_at": drnj_time.now_in_drnj_time(),
}, fail=False))
### TODO: add these users later.
### tmp_users.append(validate_document(new_user_template(), tweet_obj['user']))
if tmp_tweets:
self.application.db.insert_tweet(tmp_tweets)
# WARN: This functionality is not available for now.
# campaign_id, user_objects = extract_arguments(tmp_tweets)
# res = app_object.send_task('init_user_to_graph_offline',
# [[ campaign_id, bson.json_util.dumps(user_objects) ]],
# queue="offline_jobs")
# logger.info(str(res))
if watchlist_related:
print watchlist_related
watchlist_related = bson.json_util.loads(watchlist_related)
print watchlist_related
#self.application.db.update_watchlist(**watchlist_related)
update_task_state_in_watchlist(**watchlist_related)
else:
raise HTTPError(500, 'You tried to insert no tweets?!')
# tweets_coll.insert(tmp_tweets)
else:
tmp = []
if verbose_response:
self.write(bson.json_util.dumps({'results': tmp}))
else:
self.write(bson.json_util.dumps({'results': 'ok'}))
self.add_header('Content-Type', 'application/json')
self.finish()
except MissingArgumentError as e:
# TODO: implement logging.
raise HTTPError(500, 'You didn''t supply %s as an argument' % e.arg_name)
elif (action == 'retweets'):
try:
tweet_id = self.get_argument('tweet_id')
self.write('not implemented yet')
except MissingArgumentError as e:
# TODO: implement logging.
raise HTTPError(500, 'You didn''t supply %s as an argument' % e.arg_name)
elif (action == 'filter'):
try:
campaign_id = self.get_argument('campaign_id')
skip = self.get_argument('skip', 0)
limit = self.get_argument('limit', 100)
res_format = self.get_argument('format', 'json')
since_datetime = self.get_argument('since_datetime', -1)
until_datetime = self.get_argument('until_datetime', -1)
sort_by_datetime = self.get_argument('sort_by_datetime', 0)
produce_dbpedia_spotlight_result = self.get_argument('dbpedia_spotlight_result', 0)
tweets_coll = self.application.db.motor_column.tweets
query_string = {'campaign_id' : '%s' % campaign_id}
if since_datetime != -1:
if 'tweet.created_at' not in query_string:
query_string['tweet.created_at'] = {}
query_string['tweet.created_at']['$gte'] = float(since_datetime)
if until_datetime != -1:
if 'tweet.created_at' not in query_string:
query_string['tweet.created_at'] = {}
query_string['tweet.created_at']['$lt'] = float(until_datetime)
sort_string = []
if sort_by_datetime != 0:
sort_string = [('tweet.created_at', pymongo.ASCENDING)] # ascending
print "STARTED " + str(query_string)
cursor = tweets_coll.find(query_string)
if sort_string:
cursor = cursor.sort(sort_string)
cursor = cursor.skip(int(skip)).limit(int(limit))
# TODO: removing because of complaint:
# TypeError: if no direction is specified, key_or_list must be an instance of list
# .sort({"$natural" : 1})\
tmp = []
for x in (yield cursor.to_list(length=100)):
if int(produce_dbpedia_spotlight_result) == 1:
x['tweet']['dbpedia_spotlight_result'] = self.produce_dbpedia_spotlight_result(x['tweet']['text'])
tmp.append(x)
# tmp = [x for x in (yield cursor.to_list(length=100))]
print "ENDED " + str(query_string)
DB_TEST_VERSION = 0.2
if res_format == 'json':
self.write(bson.json_util.dumps(
{'results': tmp,
# TODO: Replace this DB_TEST_VERSION with source code
# version later
"drenaj_service_version": DB_TEST_VERSION,
# TODO: "requested_by": keywords['drnjID'],
"requested_by": "drenaj",
"campaign_id": campaign_id,
"served_at": drnj_time.now_in_drnj_time(),
'skip': int(skip),
'limit': int(limit)}))
self.add_header('Content-Type', 'application/json')
elif res_format == 'html':
env = Environment(loader=FileSystemLoader('templates'))
template = env.get_template('statuses/filter.html')
result = template.render(statuses=[x['tweet'] for x in tmp])
self.write(result)
self.finish()
except MissingArgumentError as e:
# TODO: implement logging.
raise HTTPError(500, 'You didn''t supply %s as an argument' % e.arg_name)
# db.tweets.find({'campaign_id' : 'syria'}).sort({$natural : 1}).skip(10).limit(14)
else:
# Tornado will make sure that this point will not be reached, if
# it's regexp based router works correctly
raise HTTPError(404, 'No such method')
def post(self, *args, **keywords):
"""
`view`
- params:
- user_id: user_id of the twitter user that we want to
retrieve all tweets.
`store`
- params:
- tweet_data: array of tweets in json format.
`retweets`
- params:
- tweet_id: `id` of the tweet that we want to
retrieve all retweets.
`filter`
- not implemented yet.
"""
if len(args) > 1:
(action) = args
else:
action = args[0]
print action
verbose_response = self.get_argument('verbose', '')
if (action == 'view'):
try:
# TODO: we need to get here again.
user_id = self.get_argument('user_id')
# if no user_id is supplied.
if user_id == '':
tmp = []
else:
tweets_coll = self.application.db.motor_column.tweets
# running the query
cursor = tweets_coll.find({
'tweet.user.id_str':
str(user_id),
})
tmp = [x for x in (yield cursor.to_list(length=100))]
self.write(bson.json_util.dumps({'results': tmp}))
self.add_header('Content-Type', 'application/json')
except MissingArgumentError as e:
# TODO: implement logging.
raise HTTPError(
500, 'You didn'
't supply %s as an argument' % e.arg_name)
elif (action == 'store'):
try:
tweet_data = self.get_argument('tweet_data')
campaign_id = self.get_argument('campaign_id', 'default')
watchlist_related = self.get_argument('watchlist_related', '')
if tweet_data:
#tweet_array = bson.json_util.loads(tweet_data, object_hook=self.datetime_hook)
tweet_array = bson.json_util.loads(tweet_data)
tmp_tweets = []
# TODO: Sanity check the data!
# For example, treat 'entities', 'user' specially.
DB_TEST_VERSION = 0.2
for tweet_obj in tweet_array:
tweet_obj['user']['history'] = False
tmp_tweets.append(
validate_document(
new_tweet_template(),
{
"tweet":
tweet_obj,
# TODO: Replace this DB_TEST_VERSION with source code
# version later
"drenaj_service_version":
DB_TEST_VERSION,
# TODO: "retrieved_by": keywords['drnjID'],
"retrieved_by":
"drenaj",
"campaign_id":
campaign_id,
"record_retrieved_at":
drnj_time.now_in_drnj_time(),
},
fail=False))
### TODO: add these users later.
### tmp_users.append(validate_document(new_user_template(), tweet_obj['user']))
if tmp_tweets:
self.application.db.insert_tweet(tmp_tweets)
# WARN: This functionality is not available for now.
# campaign_id, user_objects = extract_arguments(tmp_tweets)
# res = app_object.send_task('init_user_to_graph_offline',
# [[ campaign_id, bson.json_util.dumps(user_objects) ]],
# queue="offline_jobs")
# logger.info(str(res))
if watchlist_related:
print watchlist_related
watchlist_related = bson.json_util.loads(
watchlist_related)
print watchlist_related
#self.application.db.update_watchlist(**watchlist_related)
update_task_state_in_watchlist(**watchlist_related)
else:
raise HTTPError(500, 'You tried to insert no tweets?!')
# tweets_coll.insert(tmp_tweets)
else:
tmp = []
if verbose_response:
self.write(bson.json_util.dumps({'results': tmp}))
else:
self.write(bson.json_util.dumps({'results': 'ok'}))
self.add_header('Content-Type', 'application/json')
self.finish()
except MissingArgumentError as e:
# TODO: implement logging.
raise HTTPError(
500, 'You didn'
't supply %s as an argument' % e.arg_name)
elif (action == 'retweets'):
try:
tweet_id = self.get_argument('tweet_id')
self.write('not implemented yet')
except MissingArgumentError as e:
# TODO: implement logging.
raise HTTPError(
500, 'You didn'
't supply %s as an argument' % e.arg_name)
elif (action == 'filter'):
try:
campaign_id = self.get_argument('campaign_id')
skip = self.get_argument('skip', 0)
limit = self.get_argument('limit', 100)
res_format = self.get_argument('format', 'json')
since_datetime = self.get_argument('since_datetime', -1)
until_datetime = self.get_argument('until_datetime', -1)
sort_by_datetime = self.get_argument('sort_by_datetime', 0)
produce_dbpedia_spotlight_result = self.get_argument(
'dbpedia_spotlight_result', 0)
tweets_coll = self.application.db.motor_column.tweets
query_string = {'campaign_id': '%s' % campaign_id}
if since_datetime != -1:
if 'tweet.created_at' not in query_string:
query_string['tweet.created_at'] = {}
query_string['tweet.created_at']['$gte'] = float(
since_datetime)
if until_datetime != -1:
if 'tweet.created_at' not in query_string:
query_string['tweet.created_at'] = {}
query_string['tweet.created_at']['$lt'] = float(
until_datetime)
sort_string = []
if sort_by_datetime != 0:
sort_string = [('tweet.created_at', pymongo.ASCENDING)
] # ascending
print "STARTED " + str(query_string)
cursor = tweets_coll.find(query_string)
if sort_string:
cursor = cursor.sort(sort_string)
cursor = cursor.skip(int(skip)).limit(int(limit))
# TODO: removing because of complaint:
# TypeError: if no direction is specified, key_or_list must be an instance of list
# .sort({"$natural" : 1})\
tmp = []
for x in (yield cursor.to_list(length=100)):
if int(produce_dbpedia_spotlight_result) == 1:
x['tweet'][
'dbpedia_spotlight_result'] = self.produce_dbpedia_spotlight_result(
x['tweet']['text'])
tmp.append(x)
# tmp = [x for x in (yield cursor.to_list(length=100))]
print "ENDED " + str(query_string)
DB_TEST_VERSION = 0.2
if res_format == 'json':
self.write(
bson.json_util.dumps({
'results':
tmp,
# TODO: Replace this DB_TEST_VERSION with source code
# version later
"drenaj_service_version":
DB_TEST_VERSION,
# TODO: "requested_by": keywords['drnjID'],
"requested_by":
"drenaj",
"campaign_id":
campaign_id,
"served_at":
drnj_time.now_in_drnj_time(),
'skip':
int(skip),
'limit':
int(limit)
}))
self.add_header('Content-Type', 'application/json')
elif res_format == 'html':
env = Environment(loader=FileSystemLoader('templates'))
template = env.get_template('statuses/filter.html')
result = template.render(
statuses=[x['tweet'] for x in tmp])
self.write(result)
self.finish()
except MissingArgumentError as e:
# TODO: implement logging.
raise HTTPError(
500, 'You didn'
't supply %s as an argument' % e.arg_name)
# db.tweets.find({'campaign_id' : 'syria'}).sort({$natural : 1}).skip(10).limit(14)
else:
# Tornado will make sure that this point will not be reached, if
# it's regexp based router works correctly
raise HTTPError(404, 'No such method')
def prepare_hist_and_plot(self, n_tweets, users, n_bins, campaign_id):
import numpy
import matplotlib.pyplot as plot
plot_graphs = False
hist = {
'user_creation': {
'data': None,
'bins': None,
},
'user_n_tweets': {
'data': None,
'bins': None,
},
'user_n_tweets_overall': {
'data': None,
'bins': None,
},
'n_tweets': None,
'n_unique_users': None,
'n_default_profile_image': None,
'n_lower_than_threshold': None,
}
self.logger.debug("How many tweets? %d" % n_tweets)
hist['n_tweets'] = n_tweets
# TODO: abort if there are more than 200000 tweets.
if n_tweets > 200000:
return
#
# How many unique users?
#
n_unique_users = len(users)
self.logger.debug("How many unique users? %d" % n_unique_users)
hist['n_unique_users'] = n_unique_users
######
sec_title = "Histogram of user creation dates?"
#
tmp_dates = []
for x in users:
tmp_date = x['user']['created_at']
if type(tmp_date) != float:
tmp_date = py_utc_time2drnj_time(tmp_date)
tmp_dates.append(tmp_date)
# tmp_dates = [py_utc_time2drnj_time(x['user']['created_at']) for x in users]
(hist['user_creation']['data'],
hist['user_creation']['bins']) = numpy.histogram(tmp_dates,
bins=n_bins)
if plot_graphs:
bins = hist['user_creation']['bins'][:-1]
width = (hist['user_creation']['bins'][1] -
hist['user_creation']['bins'][0]) / 2
plot.bar(bins,
hist['user_creation']['data'],
width=width,
align='center')
xticklabels = [
time.strftime('%d %b %Y', time.gmtime(drnj_time2py_time(x)))
for x in bins
]
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('1.pdf', dpi=600)
#####
sec_title = "Histogram of number of tweets of each user in this campaign"
tmp_counts = [int(x['n_user_tweets']) for x in users]
#
(hist['user_n_tweets']['data'],
hist['user_n_tweets']['bins']) = numpy.histogram(tmp_counts,
bins=n_bins)
if plot_graphs:
bins = hist['user_n_tweets']['bins'][:-1]
data = hist['user_n_tweets']['data']
width = (hist['user_n_tweets']['bins'][1] -
hist['user_n_tweets']['bins'][0]) / 2
plot.bar(bins, data, width=width, align='center')
xticklabels = bins
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('2.pdf', dpi=600)
#####
sec_title = "What percentage of them used the default profile image?"
#
n_default_profile_image = 0
for u in users:
if u['user']['default_profile_image']:
n_default_profile_image += 1
hist['n_default_profile_image'] = n_default_profile_image
self.logger.debug("%s: %0.2f%%" %
(sec_title, 100 *
(float(n_default_profile_image) / n_unique_users)))
#####
sec_title = "Histogram of tweet counts of unique users"
tmp_counts = [int(x['user']['statuses_count']) for x in users]
(hist['user_n_tweets_overall']['data'],
hist['user_n_tweets_overall']['bins']) = numpy.histogram(tmp_counts,
bins=n_bins)
if plot_graphs:
bins = hist['user_n_tweets_overall']['bins'][:-1]
data = hist['user_n_tweets_overall']['data']
width = (hist['user_n_tweets_overall']['bins'][1] -
hist['user_n_tweets_overall']['bins'][0]) / 2
plot.bar(bins, data, width=width, align='center')
xticklabels = bins
plot.xticks(bins, xticklabels)
plot.title(sec_title)
#plot.show()
plot.savefig('3.pdf', dpi=600)
#
sec_title = "What percentage of them have lower than 5 tweets?"
n_lower_than_threshold = 0
for u in users:
if u['user']['statuses_count'] < 5:
n_lower_than_threshold += 1
hist['n_lower_than_threshold'] = n_lower_than_threshold
self.logger.debug("%s: %0.2f%%" %
(sec_title, 100 *
(float(n_lower_than_threshold) / n_unique_users)))
self.logger.debug(hist)
# converting numpy.array's to normal python lists.
for k in hist.keys():
if type(hist[k]) == dict:
for k2 in hist[k].keys():
if type(hist[k][k2]) == type(numpy.array([])):
hist[k][k2] = list(hist[k][k2])
hist = {
'campaign_id': campaign_id,
'histogram': hist,
'created_at': now_in_drnj_time()
}
return hist
